Chromosome Aberrations, Gene Mutations and Expression Changes, and Prognosis in Adult Acute Myeloid Leukemia

Adult acute myeloid leukemia (AML) is a very heterogeneous disease with regard to clinical features and acquired genetic alterations, both those detectable microscopically as structural and numerical chromosome aberrations, and those detected as submicroscopic gene mutations and changes in gene expression. At present, cytogenetic aberrations detected at the time of AML diagnosis constitute the most common basis for predicting clinical outcome.^1,–4 However, molecular genetic alterations are increasingly being used to refine prognosis further. In this article we briefly review major cytogenetic-risk prognostic categories and then focus on how molecular genetic findings, many of which are or likely will become therapeutic targets, contribute to prognostication of patients belonging to two of the larger cytogenetic groups: AML with a normal karyotype and core-binding factor (CBF) AML.

Acquired clonal chromosome abnormalities, i.e., a structural aberration or a trisomy observed in at least 2 and monosomy found in at least 3 metaphase cells, are detected in the pretreatment marrow of 50% to 60% of adults with de novo AML. In 10% to 20% of patients, the abnormal karyotype is complex, i.e., contains at least 3 chromosome aberrations, whereas in 40% to 50% of patients no cytogenetic abnormality can be discerned using standard banding methods.^1,–4,8,9 Recent large collaborative studies have proposed cytogenetic-risk systems classifying patients into favorable, intermediate and adverse risk groups according to the specific karyotypic findings at diagnosis (Table 1 ). Although there are some differences among these prioritization schemata, pretreatment cytogenetic findings are being used to stratify therapy.⁵ Furthermore, for patients whose pretreatment karyotype is abnormal, cytogenetic analysis has been recommended for documenting complete remission (CR),⁶ based on data showing that persistence of even 1 metaphase with an abnormality present at diagnosis results in significantly inferior cumulative incidence of relapse (CIR), disease-free survival (DFS) and overall survival (OS).⁷ 

All large cytogenetic studies of AML agree that patients with t(15;17)(q22;q12–21) have an excellent and those with t(8;21)(q22;q22) or inv(16)(p13q22)/ t(16;16)(p13;q22) a relatively favorable prognosis, whereas those with inv(3)(q21q26)/ t(3;3)(q21;q26), −7 and a complex karyotype, defined in some studies as containing at least 3 and in others at least 5 chromosome aberrations, have a poor clinical outcome (Table 1 ). Notably, the complex karyotype category usually does not include patients with t(8;21), inv(16)/t(16;16), t(15;17), or t(9;11)(p22;q23) because their prognosis has generally not been shown to be affected unfavorably by the presence of 2 or more additional chromosome aberrations.³ On the other hand, the majority of patients with −5, del(7q), −17/17p−, −18, or −20, whose prognosis is usually very poor, have a complex karyotype, which precludes evaluation of the prognostic significance of each of these chromosome aberrations individually. Interestingly, the relatively rare AML patients with del(7q) in the absence of −5/ del(5q), 3q abnormalities or a complex karyotype,¹ and those with del(5q) in a non-complex karyotype³ are classified in the intermediate-risk category.

The cytogenetic-risk classification summarized in Table 1  is based on studies that predominantly^1,3 or exclusively² included younger AML patients, below 56 or 60 years of age. Two subsequent studies analyzed the prognostic significance for pre-treatment cytogenetic findings in outcome of older AML patients, a group whose outcome is generally worse than that of younger patients. Both the United Kingdom Medical Research Council (MRC) study⁸ and the recent Cancer and Leukemia Group B (CALGB) study⁹ confirmed the value of pre-treatment karyotype as an independent prognostic factor in older patients with AML. However, the impact of some specific chromosome aberrations on clinical outcome is different in older as opposed to younger AML patients. An MRC study⁸ found the outcome of older adults with fewer than 5 abnormalities, regardless of the presence of abnormalities involving chromosomes 5, 7 and 3q, to be comparable to that of patients with a normal karyotype or “intermediate-risk” abnormalities, and therefore all these patients were included in the intermediate-risk category, leaving patients with a complex karyotype with 5 or more aberrations as the only ones in the adverse-risk category. A CALGB study confirmed that older AML patients with a complex karyotype with 5 or more aberrations have particularly poor DFS and OS, with no patient surviving 5 years after diagnosis (Table 2 ).⁹ In contrast, among older patients those with CBF AML were among those who survived the longest in both studies.^8,9 

Patients with a normal marrow karyotype are the largest cytogenetic subset of AML and are classified in the intermediate prognostic category.^1,–4 Although the risk of relapse and DFS of cytogenetically normal patients in a recent study¹⁰ were improved by postremission therapies that included 4 cycles of high- (HDAC) or intermediate-dose (IDAC) cytarabine or 1 cycle of HDAC/etoposide followed by autologous stem cell transplantation (ASCT) compared with regimens including fewer cytarabine cycles or no ASCT, not all patients benefited equally from these regimens. The likely reason for such variable response to treatment is the striking molecular heterogeneity of cytogenetically normal AML patients. During the last decade several studies have shown that the presence or absence of specific gene mutations and/or changes in gene expression have an effect on the patients’ prognosis.¹¹ At present, probably the most important prognostic factor in cytogenetically normal AML is the internal tandem duplication (ITD) of the FLT3 gene (discussed in detail in this volume by Small). Below we will review other molecular markers known to influence clinical outcome of AML patients with a normal karyotype (Table 3 ).

Heterozygous mutations in exon 12 of the nucleophosmin member 1 (NPM1) gene, resulting in abnormal cytoplasmic expression of its protein product, nucleophosmin, have recently been reported in 46% to 62% of AML patients with a normal karyotype, thus becoming the most frequent genetic alterations in this cytogenetic group of AML.^12,–16 The presence of NPM1 mutations has been associated with such pretreatment features as female sex,^13,14,16 increased bone marrow (BM) blast percentages,^13,16 LDH levels,¹³ WBC^13,–16 and platelet^13,14,16 counts, and low or absent CD34⁺ expression.^12,–14,16 Approximately 40% of patients with NPM1 mutations also harbor FLT3-ITDs, which, together with mutations in the tyrosine kinase domain of FLT3 (FLT3-TKD), are twice as common in NPM1 mutation-positive patients as in patients with wild-type NPM1.^12,–14,16 While CEBPA mutations occur with a similar frequency in patients with and without NPM1 mutations,^13,14 the partial tandem duplication (PTD) of MLL is very rare in the former.^13,14,16 

Most, but not all,¹⁵ studies analyzing either exclusively patients with a normal karyotype^13,14,16 or those classified in the intermediate-risk cytogenetic category,¹⁷ which usually contains a large proportion of cytogenetically normal patients, have demonstrated that NPM1 mutations are associated with clinical outcome. In the seminal report by Falini et al,¹² cytoplasmic localization of nucleophosmin, a feature highly correlated with the presence of NPM1 mutation, was an independent favorable prognostic factor for achievement of CR. Later studies evaluating the clinical outcome of patients with and without NPM1 mutations irrespective of the presence of other genetic rearrangements have yielded somewhat inconsistent results. In some analyses, patients with NPM1 mutations had a significantly better CR rate¹⁴ and longer event-free survival (EFS),¹⁴ relapse-free survival (RFS)¹³ and DFS¹⁶ than those with wild-type NPM1; in others, no significant difference in CR rate,¹⁵ EFS,¹⁵ RFS^14,15 and OS^13,–16 between the mutated and unmutated groups was found. However, among patients without the FLT3-ITD all studies agreed that the presence of NPM1 mutations was associated with significantly improved CR rates,^13,16 EFS,¹⁴ RFS,¹³ DFS,¹⁶ and OS.^13,14,16 In contrast, there was no major effect of NPM1 mutations on the poor outcome conferred by FLT3-ITD in patients carrying this rearrangement. The combined status of mutated NPM1 and wild-type FLT3 was also shown to be an independent favorable prognostic factor for the attainment of CR,¹³ DFS,¹⁶ EFS,¹⁴ and survival^13,16 in multivariable analyses.

Mutations of CEBPA, a gene encoding a myeloid transcription factor that plays an important role in normal granulopoiesis, have been detected in 15% to 20% of AML patients with normal cytogenetics.^15,18,19 That CEBPA mutations confer favorable prognosis was initially demonstrated in patients included in the intermediate cytogenetic-risk category^20,21 and later confirmed by studies investigating exclusively patients with a normal karyotype.^15,18,19 Compared with patients with wild-type CEBPA, those with CEBPA mutations present with higher percentages of peripheral blood (PB) blasts, lower platelet counts, less lymphadenopathy and extramedullary involvement, and are less likely to also carry FLT3-ITD, FLT3-TKD and MLL-PTD.¹⁸ While there were no significant differences in CR rates between patients with and without CEBPA mutations,^15,18,19 patients with CEBPA mutations had significantly better CRD,¹⁸ DFS,¹⁹ EFS¹⁵ and OS.^18,19 In multivariable analyses, CEBPA mutational status has added prognostic information to that provided by MLL-PTD and FLT3-ITD status, age and resistant disease after the first course of induction therapy with regard to CR duration (CRD) and by FLT3-ITD status, age and WBC count for OS.¹⁸ Similar results were obtained in another study, where CEBPA mutations, BAALC and FLT3-ITD status were prognostic for DFS; CEBPA mutations, age, BAALC and FLT3-ITD status were prognostic for OS.¹⁹ 

Preliminary data suggest that CRD of patients with N-terminal nonsense CEBPA mutations is the longest, followed by that of patients with other CEBPA mutations, including C-terminal mutations in the bZIP domain; CRD of patients with wild-type CEBPA was the shortest.¹⁸ However, pairwise comparisons did not show significant differences suggesting that larger studies are necessary to assess the prognostic role of different CEBPA mutation types.

MLL-PTD was the first molecular alteration shown to impact on clinical outcome of cytogenetically normal adults with de novo AML.²² It is detected in approximately 5% to 10% of these patients and usually involves exons 5 through 11 or, less frequently, exons 5 through 12. In 30% to 40% of patients with the MLL-PTD a FLT3-ITD is also found,¹³ whereas CEBPA¹⁸ or NPM1^13,14 mutations are rarely found together with MLL-PTD. Patients with MLL-PTD do not differ significantly from those without this mutation with regard to pretreatment characteristics. However, their CRD^23,24 is significantly shorter than that of patients without the MLL-PTD. A multivariable analysis that did not include other molecular genetic markers revealed MLL-PTD status as the only prognostically significant factor for CRD.²⁴ 

Recently, Whitman et al²⁵ have reported that the MLL wild-type allele is silenced in AML blasts harboring the MLL-PTD, most likely as a result of differential DNA methylation and histone modifications directly or indirectly regulating the promoter of the MLL wild-type allele. Importantly, the transcription of the MLL wild-type allele can be re-activated in vitro by the combination of depsipeptide, a histone deacetylase inhibitor, and decitabine (5′-aza-2′-deoxycytidine), a DNA methyltransferase inhibitor, and this leads to enhanced apoptosis.²⁵ These data suggest that MLL-PTD-positive patients might benefit from therapy that includes DNA methyltransferase and/or histone deacetylase inhibitors. However, until such molecularly targeted therapy is shown to be clinically effective in these patients, allogeneic SCT seems to be the best therapeutic approach.²⁴ 

The BAALC gene is expressed mainly in neuroectoderm-derived tissues and hematopoietic precursors.²⁶ However, high BAALC expression in PB at diagnosis was also detected in a subset of karyotypically normal adults under the age of 60 years with de novo AML, and associated with lower WBC and less frequent diagnosis of FAB M5 AML.²⁷ High BAALC expression was shown to be an independent risk factor unfavorably affecting DFS, EFS and OS.²⁷ Similar results were obtained in a second study, in which BAALC expression was measured in marrow relative to the median value of BAALC expression in marrow from healthy volunteers.¹⁹ In a subgroup analysis restricted to patients without CEBPA mutations and FLT3-ITD, high BAALC expression was again associated with inferior DFS and OS.¹⁹ 

The latest and hitherto largest study of 307 adults with AML aged 60 years or younger showed high expression of BAALC in PB to be an independent adverse prognostic factor not only for remission duration and OS, but also for resistance to initial induction chemotherapy.²⁸ On multivariable analysis, high BAALC expression and a high ratio of FLT3-ITD to wild-type FLT3 allele were the only factors predicting worse CIR and OS. Interestingly, among patients with high BAALC expression, those who underwent allogeneic SCT had a significantly lower CIR compared with patients who received autologous SCT, whereas among all patients undergoing allogeneic SCT, there was no difference in CIR between high and low BAALC expressers.²⁸ These preliminary data suggest that allogeneic SCT in first CR might overcome the adverse prognostic effect of high BAALC expression.

This is a recently identified molecular marker predicting adverse outcome in karyotypically normal AML patients. First discovered in AML patients with prognostically unfavorable complex karyotypes with cryptic amplification of chromosome,^21,29,30,ERG overexpression was also detected in a subset of cytogenetically normal AML,³⁰ raising the possibility that it might also be associated with poor prognosis in this cytogenetic subgroup of AML. This has been demonstrated in a study comprising 84 adults with de novo AML below the age of 60 years, which revealed that patients with ERG expression values in the uppermost quartile of ERG expression in pretreatment PB, who aside from having higher percentages of PB blasts did not have distinctive pretreatment characteristics, had a significantly worse CIR and OS compared with patients having lower ERG expression values.³¹ High ERG expression and FLT3-ITD were both independent prognostic factors predicting worse CIR and OS. When the analyses were restricted to the more favorable subset of patients who expressed a FLT3 wild-type allele, high ERG expression and MLL-PTD both impacted unfavorably on remission duration, whereas high ERG expression predicted shorter OS only in patients with low expression of the BAALC gene.³¹ These results require confirmation by an independent study.

Relatively little information on the clinical usefulness of gene-expression profiling in AML is available at present. A recent gene-expression profiling study revealed that most adults with AML and a normal karyotype segregated into two clusters with significantly different survival. Patients in cluster I were more frequently diagnosed with AML FAB M1 and M2 and less frequently with AML FAB M4 and M5, carried more commonly FLT3 aberrations and had shorter OS.³² However, these results require corroboration, particularly because in other gene-expression studies cytogenetically normal patients segregated into several different clusters.^33,34 The prognostic significance of the gene-expression signature identified by Bullinger et al³² has been validated recently in a patient population treated on a different protocol, with a longer follow-up (4.7 versus less than 2 years), and using a different microarray platform.³⁵ Moreover, this study has increased the clinical applicability of the signature because a class prediction algorithm was created that allows for the prediction of outcome for subsequent individual patients. This signature-based classifier identified groups with differences in DFS and OS.³⁵ However, the outcome classifier was strongly associated with the FLT3-ITD, which might explain the prognostic significance of the signature. Additionally, the classifier for dichotomized outcome classes had only moderate predictive accuracy, with OS and DFS of around 60% of the patients being correctly predicted. Nevertheless, the classifier showed some ability to identify a subset of patients with wild-type FLT3 whose outcome is poor.³⁵ Future studies should aim at identifying other classifiers that might predict outcome for individual AML patients with a normal karyotype with greater precision.

CBF AML is a relatively frequent subtype of adult de novo AML, with t(8;21) being detected in 7% and inv(16)/ t(16;16) in 8% of patients.³ Nevertheless, few studies including over 100 patients have assessed the clinical characteristics of t(8;21) and inv(16) AML,^36,–40 and only three compared the two cytogenetic subsets of CBF AML directly.^38,–40 Such a comparison is important because CBF AML patients with t(8;21) are often grouped together with those harboring inv(16) in clinical studies based on similar, relatively favorable prognoses, which can be improved by postremission treatment with 3 or 4 cycles of IDAC or HDAC as opposed to 1 cycle,³⁹ and the fact that t(8;21) and inv(16) are related at the molecular level because they disrupt the α and β subunits of CBF, respectively.

In addition to a characteristic FAB M4Eo morphology that distinguishes inv(16)/t(16;16)-positive patients from those with other karyotypes, patients with inv(16)/t(16;16) differ from t(8;21) patients with regard to several other pre-treatment features. They are more frequently white and less frequently African American,^39,40 have higher WBCs,^38,–40 percentages of PB and BM blasts,³⁹ more often extramedullary involvement,^38,39 in particular lymphadenopathy,^38,39 splenomegaly,³⁹ gingival hypertrophy³⁹ and skin/mucosa involvement,³⁸ and characteristic cytogenetic features. Approximately two thirds of patients with inv(16)/t(16;16) have this rearrangement as a sole chromosome abnormality, whereas sole t(8;21) is detected in only about 30% of patients with this translocation.^38,–40 The pattern of secondary chromosome aberrations is also different, with loss of a sex chromosome (−Y in men and −X in women) and del(9q) being the most frequent secondary chromosome aberrations in t(8;21)-positive patients, and +22, +8, del(7q) and +21 in inv(16)/t(16;16)-positive patients.^38,–40 Importantly, marked molecular genetic differences between the two cytogenetic groups have been revealed by microarray gene-expression studies that can differentiate patients with inv(16)/t(16;16) from those with t(8;21) with greater than 95% accuracy based on their gene-expression signatures.⁴¹ 

Both cytogenetic subgroups have very similar, relatively high CR rates in the range of 85% to 89%,^38,–40 and in both, the probability of CR attainment is negatively affected by lower platelet counts.³⁹ However, lower CR probability was associated with hepatomegaly only in inv(16)/ t(16;16) patients, and with higher BM blasts and, unexpectedly, nonwhite race exclusively in the t(8;21) group. Nonwhite patients with t(8;21) had 5.7 times the odds of not achieving a CR compared with the corresponding white population. The reasons for such a difference are unknown, but because the patients were treated on the same protocols the difference might be related to a true biologic diversity resulting in more resistant disease and/or higher susceptibility to drug toxicity of nonwhite patients with t(8;21).³⁹ 

Univariable analyses did not reveal significant differences in relapse risk or OS between inv(16)/t(16;16) and t(8;21) groups.^38,–40 However, after adjusting for age, log[WBC], and log[platelets], patients with t(8;21) were found to have a significantly shorter OS than those with inv(16)/t(16;16).³⁹ This difference is likely related to a dissimilar response to salvage therapy, since patients with t(8;21) had a significantly shorter survival after relapse compared with inv(16)/t(16;16) patients in three large, independent studies.^38,–40 These studies identified additional cytogenetic prognostic factors differentiating the two cytogenetic subsets of CBF AML. Among patients with inv(16)/t(16;16), those who harbored +22 as a secondary abnormality were found to have a significantly lower CIR compared with patients with inv(16)/t(16;16) as a sole abnormality in the CALGB study³⁹ and longer RFS than patients without +22 in the German Acute Myeloid Leukemia Intergroup study.³⁸ In the latter study,³⁸ loss of the Y chromosome (−Y) in male patients with t(8;21) was associated with a shorter OS, although this result could not be corroborated in the CALGB series.³⁹ Instead, Marcucci et al³⁹ observed a possible interaction between secondary chromosome abnormalities and race in patients with t(8;21). Nonwhites with secondary cytogenetic abnormalities other than del(9q) had shorter OS than nonwhites who had an isolated t(8;21) or a secondary del(9q). In contrast, among white patients, neither del(9q) nor other secondary aberrations, including −Y in men, influenced the patients’ survival.

Because a substantial proportion of patients with CBF AML carry mutations in the KIT gene (Table 4 ), recent studies examined prognostic significance of these mutations. As shown in Table 4 , various KIT mutations analyzed as one group and, more specifically, mutations in exon 17, which encodes the activation loop in the kinase domain of KIT, have been associated with inferior OS,^42,–44 EFS,^43,44 RFS,⁴⁴ relapse incidence⁴² and CIR⁴⁵ in patients with t(8;21). The prognostic significance of KIT mutations in patients with inv(16)/t(16;16) has been less firmly established. Care et al⁴⁶ found KIT mutations in exon 8 to unfavorably affect the relapse rate (but not OS), whereas two smaller studies did not detect any prognostic impact of KIT mutations.^42,44 In a recent CALGB study,⁴⁵ performed on a relatively large number of similarly treated patients with inv(16), the presence of all KIT mutations bestowed a higher CIR. Notably, subset analyses revealed that the difference in CIR was mainly caused by the presence of mutations in exon 17 of KIT. Patients with these mutations had over six times higher CIR than those with wild-type KIT. In multivariable analyses, KIT mutations, both those in exon 17 and exon 8, impacted adversely on OS after adjusting for sex.⁴⁵ These results await independent confirmation.

The discovery of prognostically relevant KIT mutations in CBF AML is important because it has potential therapeutic significance. Mutations in the KIT gene have been previously shown to constitute potential targets for tyrosine kinase (TK) inhibitors. Importantly, specific TK inhibitors are active against particular KIT mutations, e.g., imatinib is active against exon 17 mutations involving N822 in the activation loop, or variants of mutations in exon 8, but not against D816 mutations in the exon 17 activation loop. On the other hand, exon 17 D816 mutations can be successfully targeted with other TK inhibitors, such as PKC412 and dasatinib. Hence, it is critical to establish the exact type of KIT mutation in each case.⁴⁵ The efficacy of TK inhibitors as part of therapy administered to patients with CBF AML should be investigated in future clinical trials.

It has been well established that AML is a very heterogeneous disease at both the cytogenetic and molecular genetic levels. Because cytogenetic findings are among the most important prognostic factors, cytogenetic analysis of BM is now mandatory in the diagnostic workup of newly diagnosed patients with AML.⁴⁷ Results of this analysis are also used for selecting therapy. The presence of fusion genes characteristic of CBF AML, acute promyelocytic leukemia with t(15;17) and AML with rearrangements of band 11q23 can also be detected by RT-PCR and fluorescence in situ hybridization (FISH). These methods are especially valuable in patients with variants of typical cytogenetic aberrations and in patients suspected of carrying cryptic rearrangements, e.g., those with FAB M4Eo, M3 or M3v marrow morphology but without, respectively, inv(16)/t(16;16) and t(15;17) on standard cytogenetic examination. Cytogenetic, FISH and molecular analyses for the more common recurring translocations are readily available.⁴⁸ 

The clinical significance of some of the molecular genetic tests discussed in this review, e.g., microarray gene-expression profiling and testing for ERG gene expression, is not yet firmly established and requires further study before such testing can be translated into clinical practice. Moreover, although testing for FLT3 mutations in younger patients with de novo AML is now recommended by the Practice Guidelines in Oncology,⁴⁷ testing for FLT3-ITD and for the other molecular markers is available mostly at the large university centers and performed as part of clinical trials. Because two or more genetic alterations can be present simultaneously in leukemic blasts of the same patient, studies that would test all currently known molecular alterations simultaneously to determine the relative impact of each of them are in progress, as are clinical trials of FLT3 inhibitors used in combination with chemotherapy.¹¹